Identifying DEmRNAs, DEmiRNAs, and DElncRNAs in HCC
The RNA expression profiles of patients with HCC and corresponding clinical information were downloaded from the TCGA database using the Data Transfer Tool. We used EdgeR to identify significantly DEmRNAs, DEmiRNAs, and DElncRNAs between HCC tissues and adjacent normal tissues. This identified a total of 1992 DEmRNAs, 122 DEmiRNAs, and 1082 DElncRNAs. More specifically, there were 1787 (89.7%) up-regulated and 205 (10.3%) down-regulated DEmRNAs (Table S1), 119 (97.5%) up-regulated and 3 (2.5%) down-regulated DEmiRNAs (Table S2), and 58 (5.4%) down-regulated DElncRNAs and 1024 (94.6%) up-regulated DElncRNAs identified (Table S3). A heat map demonstrating the complete linkage clustering of DEmRNAs, DEmiRNAs, and DElncRNAs is shown in Fig. 1.
Predictions of mRNAs and lncRNAs targeted by miRNAs
Next, we predicted the mRNAs and lncRNAs that were targeted by miRNAs, focusing on the relationship the 122 differentially expressed miRNAs and 1082 differentially expressed lncRNAs above. Only 15 of 122 differentially expressed miRNAs were predicted to target 76 of 1082 differentially expressed lncRNAs (table1). The relationships between these 15 differentially expressed lncRNA-targeting miRNAs were used to predict the targeted mRNAs using Targetscan, miRTarBase, and miRDB. Then, 15 HCC-specific miRNAs were predicted to target the 35 mRNAs (table2).
Table1. 15 DEmiRNAs interact with the 76 DElncRNAs
miRNA
|
LncRNA
|
hsa-mir-424
|
MYLK-AS1 C2orf48 CCDC13-AS1 AL033381.1 AP002478.1 FAM87A WT1-AS TCL6 AC087392.1 AC006305.1 AC016773.1 LINC00473 WARS2-IT1 SFTA1P LINC00355 LINC00200 LINC00160 DLX6-AS1 BPESC1 DSCR10 TSPEAR-AS1 GPC6-AS1 CLRN1-AS1 PART1 HOTTIP GDNF-AS1 PVT1 RMST LINC00485
|
hsa-mir-519d
|
C2orf48 AP002478.1 FAM87A WT1-AS LINC00308 LINC00221 TCL6 AC087392.1 AC061975.6 AL512652.1 TDRG1 AL359878.1 AC006305.1 HOTAIR LINC00200 DLX6-AS1 LINC00462 SRGAP3-AS4 HOTTIP PVT1 GRM5-AS1 RMST AC040173.1
|
hsa-mir-217
|
AL357153.1 AC024563.1 WT1-AS LINC00221 TCL6 AC016773.1 HOTAIR LINC00200 MIR137HG LINC00494 CLRN1-AS1 MYLK-AS1 AP000553.1 CRNDE PVT1 AC040173.1 NOVA1-AS1
|
hsa-mir-182
|
AP002478.1 WT1-AS LINC0022 TCL6 AC006305.1 AL163952.1 ZNF385D-AS2 LINC00114 SFTA1P ERVMER61-1 MIR137HG ERVH48-1 LINC00494 MYLK-AS1 AC012640.1 AC073352.1 GRM5-AS1 RMST AC040173.1
|
hsa-mir-506
|
AL033381.1 FAM87A LINC00221 LINC00501 AL359878.1 AC006305.1 LINC00355 HOTAIR LINC00200 SACS-AS1 ERVMER61-1 DLX6-AS1 BPESC1 LINC00316 MYLK-AS1 HOTTIP PVT1 RMST
|
hsa-mir-183
|
C2orf48 AL357153.1 AC024563.1 TCL6 LINC00501 LINC00392 LINC00200 CRNDE PVT1 AC040173.1
|
hsa-mir-96
|
FAM87A WT1-AS LINC00221 TCL6 AL163952.1 LINC00488 ZNF385D-AS2 LINC00114 ERVMER61-1 ERVH48-1 AC073352.1 GRM5-AS1 RMST AC040173.1
|
hsa-mir-373
|
C2orf48 AC009065.1 AP002478.1 C10orf91 WT1-AS LINC00221 TCL6 AC087392.1 AC061975.6 AL359878.1 SACS-AS1 DLX6-AS1 LINC00462 LINC00494 LINC00322 HOTTIP PVT1 GRM5-AS1 LINC00485
|
hsa-mir-372
|
C2orf48 AC009065.1 AP002478.1 C10orf91 WT1-AS LINC00221 TCL6 AC087392.1 AC061975.6 AL359878.1 SACS-AS1 DLX6-AS1 LINC00462 LINC00494 LINC00322 HOTTIP PVT1 GRM5-AS1 LINC00485
|
hsa-mir-141
|
PART1 CCDC13-AS1 AL357153.1 AC024563.1 FAM87A WT1-AS LINC00308 AL512652.1 AL359878.1 LINC00355 AL713998.1 CCDC26 DLX6-AS1 BPESC1 ERVH48-1 AC009121.1 MYLK-AS1 AC114489.1 HOTTIP LINC00485
|
hsa-mir-137
|
LINC00308 TCL6 AC006305.1 AC073263.1 CLDN10-AS1 HTR2A-AS1 AL713998.1 ERVH48-1 GPC6-AS1 CLRN1-AS1 AC012640.1 HOTTIP RMST
|
hsa-mir-205
|
PART1 CCDC13-AS1 AP002478.1 FAM87A LINC00308 TCL6 AL512652.1 AL163952.1 LINC00488 ZNF385D-AS1 LINC00351 CCDC26 SACS-AS1 ERVMER61-1 BPESC1 GPC6-AS1 CLRN1-AS1 MYLK-AS1 AC012640.1 HOTTIP AC011453.1 CRNDE PVT1 GRM5-AS1 RMST LINC00485
|
hsa-mir-216a
|
C2orf48 AP002478.1 TCL6 AC087392.1 AL359878.1 AC006305.1 LINC00488 AL357060.1 AC016773.1 LINC00114 WARS2-IT1 SFTA1P HOTAIR LINC00200 DLX6-AS1 BPESC1 AL589947.1 CLRN1-AS1 MYLK-AS1 HOTTIP GDNF-AS1 PVT1 NOVA1-AS1 LINC00485 LINC00519
|
hsa-mir-184
|
AP002478.1 ERVH48-1 TSPEAR-AS1 HOTTIP LINC00491
|
hsa-mir-216b
|
FAM87A WT1-AS TCL6 AL163952.1 LINC00488 WARS2-IT1 SFTA1P HOTAIR LINC00200 DLX6-AS1 BPESC1 CLRN1-AS1 AC012640.1 CRNDE PVT1 LINC00491 GRM5-AS1 NOVA1-AS1
|
Table2. 15 DEmiRNAs interact with the 35 DEmRNAs
miRNA
|
mRNA
|
hsa-mir-424
|
CCNE1 CDC25A AXIN2 E2F7 CEP55 KIF23 CBX2 HOXA10 GNAL HOXA3 ITGA2 CLSPN CPEB3
|
hsa-mir-519d
|
KIF23 E2F2 NETO2 E2F1 ELAVL2 POLQ RRM2 SALL3 ACSL4
|
hsa-mir-217
|
EZH2 DACH1
|
hsa-mir-182
|
NPTX1 FOXF2 HOXA9
|
hsa-mir-506
|
LRRC1 ZWINT
|
hsa-mir-183
|
GLUL CCNB1
|
hsa-mir-96
|
PROK2
|
hsa-mir-373
|
ELAVL2 SLC7A11 PBK
|
hsa-mir-372
|
SLC7A11 ELAVL2
|
hsa-mir-141
|
ELAVL2 EPHA2
|
hsa-mir-137
|
PTGS2
|
hsa-mir-205
|
ACSL4
|
hsa-mir-216a
|
NOM1 LAMC1 PAK1 FAM117B TGFBR2 SP4 MYLIP GPBP1 MTO1 TWISTNB OXGR1
|
hsa-mir-184
|
LRRC8A
|
hsa-mir-216b
|
PPP2CB COL4A4 TM9SF3 ZDHHC9 SMAD1 TPM3 KLF12 CCDC65 DNAJB9 ZNF566 C11orf57 AKIP1 SOCS6 ARL6IP1 FZD5 MCM4
|
Construction of the ceRNA network
To reveal how lncRNA may mediate transcription in HCC by affecting mRNA and miRNA binding, a ceRNA network based on the lists of DElncRNAs, DEmiRNAs, and DEmRNAs was constructed and visualized using Cytoscape software. As shown in Fig. 2, the lncRNA-miRNA-mRNA network was comprised of 15 miRNA nodes, 35 mRNA nodes, 76lncRNA nodes.
Survival analysis with differentially expressed lncRNAs and mRNAs
To investigate the relationship between the differentially expressed lncRNAs and mRNAs and the prognosis of HCC patients, the Kaplan–Meier method was used to analyze the relationship between the differential expression of 76 lncRNAs and 35 mRNAs and the overall survival rate in HCC patients. The most significant 10 of 77 differentially expressed lncRNAs were associated with the prognosis in HCC: MYLK-AS1, AL163952.1, ERVMER61-1, WARS2-IT1, AC073352.1, HTR2A-AS1, CLRN1-AS1, AL359878.1, C10orf91, AP002478.1 (log-rank P < 0.05) (Fig. 3). The most significant 10 of 35 differentially expressed mRNAs were linked to the prognosis in HCC: CCNE1, NPTX1, CPEB3, E2F2, RRM2, KIF23, E2F7, CDC25A CCNB1, E2F1 (log-rank P < 0.05) (Fig. 4).
Functional enrichment analysis based on mRNAs
To establish context of ceRNA network, we inferred the roles of each lncRNA based on the functions of connected mRNAs. lncRNAs were typically central and connected to one or more mRNAs in the network. The top 16 highly enriched GO terms of biological process (BP), cellular component (CC) and molecular function (MF). GO analysis was conducted to ascertain the signaling cascade that the 25 genes participate(Fig. 5). Finally, KEGG pathway analysis revealed that 9 pathways were significantly enriched, particularly mRNAs involved in cancer(Fig. 6).
Prognostic value of MYLK-AS1—has-mir-424—CCNE1 ceRNA network
ROC curves were constructed to evaluate the sensitivity and specificity of MYLK-AS1—has-mir-424—CCNE1 ceRNA network for prediction of HCC diagnosis. The area under the ROC curve of MYLK-AS1, has-mir-424 and CCNE1 was 0.91, 0.97 and 0.95 respectively. (Fig. 7A-C). To investigate whether MYLK-AS1 correlated with CCNE1 in HCC, we performed expression analysis and found a positive correlation between MYLK-AS1 and CCNE1 expression in HCC using TCGA database [19](Fig.7D)
Differential expression of MYLK-AS1 in HCC cells and tissues
In our study, Our results showed that MYLK-AS1 expression was significantly increased in two HCC cell lines (Huh7, HepG2) compared with that in normal HCC epithelial cell line LO2 (Fig.8A).MYLK-AS1 expression was higher in HCC tissues than adjacent normal tissues by human HCC tissue and adjacent normal tissue microarrays. (Fig.8B, C). In addition, MYLK-AS1 expression was further analysed according to patients’ clinical pathological features. Higher expression of MYLK-AS1 correlated with larger tumour size, advanced TNM stage (Fig.8D, E). The area under the ROC curve of MYLK-AS1 was 0.75(Fig.8F). Kaplan‐Meier survival analysis and log‐rank tests showed that higher MYLK-AS1 expression was associated with shorter survival time (Fig. 8G). Collectively, these findings indicate that MYLK-AS1 is a potential biomarker for diagnosis and prognosis in HCC.
MYLK-AS1 silencing inhibits HCC cells proliferation,migration and invasion
To examine the biological functions of MYLK-AS1, the si-MYLK-AS1 1#, si- MYLK-AS1 2#, si-MYLK-AS1 3#, or si negative control (NC) were separately transfected into Huh7 and HepG2 cells. We found that MYLK-AS1 expression was remarkably reduced in si-MYLK-AS1 3# transfected into HCC cells compared with si NC (Fig. 9A). The CCK8 assay and plate clone formation assay was used to determine the role of MYLK-AS1 in cell growth. In Huh7 and HepG2 cells, MYLK-AS1 knockdown led to reduced cell proliferation compared with that observed in the si NC cells (Fig. 9B, C). Furthermore, in the wound healing assay, knockdown of MYLK-AS1 contributed to slower scratch healing (Fig. 9D). Transwell assay revealed that the invasion ability of cells in which MYLK-AS1 was silenced was suppressed compared with that of si NC (Fig.9E). These results reported that MYLK-AS1 could promote the proliferation and metastasis of HCC cells.
MYLK-AS1 functions as a ceRNA and sponges miR-424-5p in HCC cells.
To investigate the molecular mechanism by which MYLK-AS1 acting as ceRNAs. First of all, MYLK-AS1 knockdown significantly increased the expression level of miR-424-5p (Fig. 10A), This finding suggested that MYLK-AS1 may function as a ceRNA of miRNAs. To determine this hypothesis, We then performed dual luciferase reporter assays to confirm the prediction analysis. HEK293T cells were transfected with a luciferase plasmid harboring the sequence of MYLK-AS1 together with plasmids encoding the miRNAs or a control sequence. We found that miR-424-5p could suppress MYLK-AS1-driven luciferase activity, and the suppression ability of miR-424-5p is stronger (Fig. 10B). To determine whether miR-424-5p functions as a tumor suppressor in HCC cells, we transfected Huh7 and HepG2 cells with miR-424-5p mimic or inhibitor (Fig. 10C). Real-time PCR analysis confirmed that the expression of MYLK-AS1 was lower in miR-424-5p mimic compared with mimic control, the expression of MYLK-AS1 was higher in miR-424-5p inhibitor compared with inhibitor control(Fig. 10D). We then performed CCK-8 and colony formation found that cell proliferation and colony formation ability were significantly reduced by overexpression of miR-424-5p (Fig. 10E,F)and significantly enhanced by silencing of miR-424-5p expression (Fig.11A,B). the wound healing assay and transwell assay found that cell migration and invasive ability were significantly reduced by overexpression of miR-424-5p (Fig. 10G,H) and significantly enhanced by silencing of miR-424-5p expression (Fig.11C,D)
CCNE1 is a miR-424-5p target gene and is indirectly regulated by MYLK-AS1
To determine the ceRNA network between MYLK-AS1, miR-424-5p and its targets in HCC. We found that knockdown of MYLK-AS1 also significantly reduced CCNE1 mRNA and protein levels in Huh7 and HepG2 cells (Fig. 12A). To determine whether CCNE1 is regulated by miR-424-5p liver cancer cells, we measured CCNE1 mRNA and protein levels when miR-424-5p was over-expressed or inhibited in Huh7 and HepG2 cells. We found that CCNE1 mRNA and protein levels were significantly decreased by miR-424-5p overexpression (Fig. 12B), in contrast, CCNE1 mRNA and protein levels were significantly increased by miR-424-5p inhibition (Fig. 12C). Nest, we performed luciferase reporter assays driven by the wild-type 3’ UTR sequence of CCNE1, which contains the predicted miR-424-5p binding site (wt- CCNE1), or mutant constructs containing a mutation in the miR-424-5p-binding sites (mut- CCNE1). These plasmids were co-transfected into HEK293T cells together with miRNA mimic control or miR-424-5p mimic. The results showed that wt- CCNE1-driven luciferase expression was significantly reduced by co-transfection with the miR-424-5p mimic compared with the control, but this repression was abolished by mutation of the putative miR-424-5p-binding site in the CCNE1 3’UTR (Fig. 12D). Taken together, these results indicate that miR-424-5p regulates CCNE1 expression in liver cancer cells by directly binding to the predicted site in the 3’ UTR of CCNE1 mRNA.
miR-424-5p plays a role in the relationship between MYLK-AS1and CCNE1
To determine whether miR-424-5p is involved in mediating the effects of MYLK-AS1 in HCC cells, Huh7 and HepG2 cells were co-transfected with si- MYLK-AS1 3# and miR-424-5p inhibitor. Notably, In addition, proliferation and colony forming assays revealed that inhibition of the miR-424-5p promoted the proliferation of Huh7 and HepG2 cells, and this effect was partly reversed by co-transfection with si- MYLK-AS1 3# (Fig. 13A and B), the wound healing assay and transwell invasion assay found that was partially rescued by co-transfection with miR-424-5p inhibitor(Fig. 13C and D). To determine whether miR-424-5p plays a role in the relationship between si-MYLK-AS1 3# and CCNE1, we examined cells co-transfected with si-MYLK-AS1 3# and the miR-424-5p inhibitor. Indeed, the suppression of CCNE1 protein levels induced by si-MYLK-AS1 3# was effectively reversed by the miR-424-5p inhibitor(Fig. 13E). Collectively, these data suggest that MYLK-AS1 modulates the expression of CCNE1 by post-transcriptional regulation of miR-424-5p.